sign in sign up
<<
Home , Archaeophyte, Arundo donax

Origin of the invasive donax (): a trans-Asian expedition in herbaria Laurent Hardion, Regine Verlaque, Kristin Saltonstall, Agathe Leriche, Bruno Vila

To cite this version:

Laurent Hardion, Regine Verlaque, Kristin Saltonstall, Agathe Leriche, Bruno Vila. Origin of the invasive Arundo donax (Poaceae): a trans-Asian expedition in herbaria. Annals of , Oxford University Press (OUP), 2014, 114 (3), pp.455-462. ￿10.1093/aob/mcu143￿. ￿hal-01164063￿

HAL Id: hal-01164063 https://hal.archives-ouvertes.fr/hal-01164063 Submitted on 7 May 2018

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Annals of Botany 114: 455–462, 2014 doi:10.1093/aob/mcu143, available online at www.aob.oxfordjournals.org

Origin of the invasive Arundo donax (Poaceae): a trans-Asian expedition in herbaria

Laurent Hardion1,*, Re´gine Verlaque1, Kristin Saltonstall2, Agathe Leriche1 and Bruno Vila1 1Institut Me´diterrane´en de Biodiversite´ et d’Ecologie, UMR CNRS IRD, Aix Marseille Universite´, 13331 Marseille, France and 2Smithsonian Tropical Research Institute, Apartado 0843-03092, Panama * For correspondence. E-mail [email protected]

Received: 7 March 2014 Returned for revision: 13 May 2014 Accepted: 2 June 2014 Published electronically: 31 July 2014

† Background and Aims The hypothesis of an ancient introduction, i.e. archaeophyte origin, is one of the most chal- lenging questions in phylogeography. Arundo donax (Poaceae) is currently considered to be one of the worst globally, but it has also been widely utilzed by man across Eurasia for millennia. Despite a lack of phylogen- etic data, recent literature has often speculated on its introduction to the Mediterranean region. † Methods This study tests the hypothesis of its ancient introduction from to the Mediterranean by using DNA sequencing and morphometric analysis on 127 specimens collected across sub-tropical Eurasia. In addition, a bioclimatic species distribution model calibrated on 1221 Mediterranean localities was used to identify similar ecological niches in Asia. † Key Results Despite analysis of several plastid DNA hypervariable sites and the identification of 13 haplotypes, A. donax was represented by a single haplotype from the Mediterranean to the Middle East. This haplotype is shared with invasive samples worldwide, and its nearest phylogenetic relatives are located in the Middle East. Morphometric data characterized this invasive clone by a robust morphotype distinguishable from all other Asian samples. The ecological niche modelling designated the southern Caspian Sea, southern Iran and the Indus Valley as the most suitable regions of origin in Asia for the invasive clone of A. donax. † Conclusions Using an integrative approach, an ancient dispersion of this robust, polyploid and non-fruiting clone is hypothesized from the Middle East to the west, leading to its invasion throughout the .

Key words: Arundo donax, Poaceae, invasive species, archaeophyte, herbarium specimens, phylogeography, domesticated species, Mediterranean, Asia, morphometry, clonal species, , giant , giant .

INTRODUCTION (Pysek et al., 2004a). In addition, several archaeophytes have become neophytes, following secondary introductions from Human activities have disrupted the distribution ranges and dis- Eurasia and Africa to the Americas (La Sorte and Pysek, 2009). persal barriers of many organisms, introducing them both delib- However, many assumptions of ancient introductions remain hy- erately (e.g. for food, gardening, construction or erosion control) potheses needing validation by further investigation (Preston and accidentally around the globe (Elton, 1958; Pysek and et al., 2004). For many archaeophytes, the literature cannot desig- Richardson,2006). Some species have become invasive and con- natea‘native’or‘introduced’status,apositiontermed‘cryptogen- stitute a major threat to global biodiversity by competing with ic species’ (Carlton, 1996). Thus, the origins of archeophytes are native species for ecosystem resources (Shea and Chesson, often difficult to identify, even using phylogenetic investigations. 2002). These alien taxa can also cause significant damage and Nevertheless, several distinctive generalizations can be used to economic losses to human activities (Pimentel et al., 2005). To help identify ancient introductions: (1) many archaeophytes control and reduce their impacts, specific knowledge is needed were introduced with translocations, and their affinities for regarding not only their biology and ecology (Dandelot et al., agricultural or human-disturbed habitats are always noticeable 2005; Blight et al., 2012), but also their genetic and taxonomic despite millennia of occurrence (Pysek et al., 2005); (2) due to delimitation (Saltonstall, 2002; Verlaque et al., 2011), and their ancient presence, archaeophytes are generally more wide- their geographical origins (Lafuma et al., 2003; Verlaque spread than neophytes (Preston et al., 2004; Pysek et al., et al., 2003). Indeed, the designation of weed origin can be ex- 2004b); and (3) in their introduced ranges, alien taxa often tremely useful from a management perspective as it can assist contain lower genetic diversity than in their native area due to with limiting introduction vectors, studying the adaptive poten- founder effects and genetic drift following their introduction tial of invasive species in their native environments and inform- (Dlugosch and Parker, 2008). However, this genetic pattern may ing biological control efforts (Clay, 2003). be disrupted by recurrent introductions and admixture of invasive Among alien species, the literature distinguishes archaeo- populations or by hybridization and introgression with native taxa phytes, i.e. species introduced between the beginning of (Baumel et al., 2002; Suehs et al., 2004). Neolithic and the discovery of the Americas (approx. Among invasive species, Arundo donax (Poaceae) represents 1500 AD), from neophytes, i.e. taxa spread after 1500 AD one of the most aggressive taxa in sub-tropical andtemperate

# The Author 2014. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For Permissions, please email: [email protected] 456 Hardion et al. — Origin of the invasive Arundo donax (Poaceae) . This giant cane is a tall (up to 6 m), rhizomatous grass order to extract the most information from these specimens, we considered invasive in many warm regions, including Oceania, set up a three-tiered integrative approach: (1) phylogenetics, Africa and the Americas (Herrera and Dudley, 2003)whereit using sequencing of plastid DNA micro-/minisatellites and sub- has invaded mainly riparian areas via clonal growth and dispersal stitutions; (2) morphometry, including stomatal length as an es- (Wijte et al., 2005). Described from and south-eastern timation of ploidy; and (3) ecology, using ecological niche France by Linneaus in 1753, A. donax has generally been consid- modelling. eredasnativetosub-tropicalEurasia(includingtheMediterranean Basin, MiddleEast,northernIndiaandEast Asia),butthe originof invasive populationsremains unknown. Asan emblematic species MATERIALS AND METHODS of Mediterranean landscapes, this robust reed mainly occurs in Plant collections human-disturbed habitats, including agrosystems, wastelands, and riparian and coastal zones. Since antiquity, giant cane has had mul- The Eurasian sampling of Arundo donax was divided into three tiplehuman uses, such as agriculture, fodder, construction, weap- inclusive datasets, according to the markers used: (1) 57 well- onry, fishing, hunting, music, erosion control, medicine and fuel preserved herbarium specimens suitable for molecular analyses; (Perdue, 1958; Ta¨ckolm and Drar, 1973; Postgate, 1980). In add- (2) a broader collection of 127 herbarium specimens with ition, this species is one of the most promising crops for and panicle used for morphometric measurements (Supplementary production, leading to an increasing literature on its Data Table S1); and (3) a large dataset of 3429 occurrences for genetic characterization (Pilu et al., 2014; Sablok et al., 2014). A. donax in the Mediterranean region. Provided by B, BM, E, G, Previous work on A. donax has revealed low genetic diversity K, MARS, P and W Herbaria, these valuable specimens were col- across broad geographical areas. Ahmad et al. (2008) tested 185 lected from the 18th century to the present, in particular by some putative clones of A. donax from the southern USA using SRAP- pioneering botanists in Asia such as V. Jacquemont (1757– (sequence-related amplified polymorphism) and transposable 1836),J.G.Ko¨nig (1728–1785), T. Thomson (1817–1878) and element (TE)-based molecular markers, and found only one J. S. Gamble (1847–1925). We also included three samples of in- widespread genotype that was similar to four populations from vasive A. donax collected from New Caledonia (Oceania), Peru southern France. Further, Hardion et al. (2012) used AFLP (amp- () and Arizona, USA (), and lified fragment length polymorphism) fingerprints to show the seven samples of the Taiwanese endemic Arundo formosana occurrence of only one clone among 16 localities around the (2n ¼ 12xapprox.72;Hardion etal., 2013) which is aclose relative Mediterranean Basin. This lack of genetic diversity calls into of A. donax (Hardion et al., 2012). question the native status of A. donax in the Mediterranean, and redirects research of native populations toward Asia. Using DNA extraction, sequencing and phylogeographic analysis AFLP and inter-simple sequence repeat (ISSR) markers on 67 Mediterranean and ten Asian herbarium specimens, Mariani DNA extractions were performed on samples collected after et al. (2010) showed support for a monophyletic origin of 1930 to avoid unnecessary destruction of older specimensand se- A. donax in Asia. This eastern origin is also supported by the quencing of highly degraded DNA (Telle and Thines, 2008). botanical literature, which reports lower cytotypes (2n ¼ 12x About 50 mg of leaves were mechanically ground after treatment approx. 72) in Thailand (Larsen, 1963), (Christopher and with liquid . Total DNA was extracted following Doyle Abraham, 1971; Mehra and Kalia, 1975) and Uzbekistan and Doyle (1987). DNA concentrations were estimated using a (Bochantseva, 1972) than in the Mediterranean region (2n ¼ Biophotometer (Eppendorf, Germany) and diluted to 50 ng mL–1. 18x approx. 108; Pizzolongo, 1962; Gorenflot et al., 1972; Plastid DNA diversity was screened on five intergenic spacers: Hardion et al., 2011, 2013; Bucci et al., 2013). Further, although trnT-trnL (Taberlet et al., 1991), trnCF-rpoB, psaA-ORF170, seedproduction has not been detected in the Mediterranean or the rbcL-psaI and trnS(GCU)-psbD (Saltonstall, 2001). Polymerase USA (Johnson et al., 2006; Hardion et al., 2012), caryopses have chain reactions (PCRs) were performed in 50 mL volumes contain- been found in Iran and Afghanistan (Bor, 1970), China and ing 1 × PCR buffer [10 mM Tris–HCl, 50 mM KCl, 0.001 % (w/v) (Brach and Song, 2006). The non-fruiting character of gelatin], 1.5mM MgCl2,2.5mM each dNTP, 40 pmol each primer, Mediterranean populations has been attributed to their high 0.1 mg mL–1bovine serum albumin (BSA) and 2.5UofTaq poly- ploidy level which may disrupt meiotic processes (Balogh merase (Q-Biogen, Illkirch, France). The thermal cycling profile et al., 2012). was programmed on a PTC-200 Gradient Thermal Cycler (MJ The widespread ruderal distribution of a single non-fruiting Research, Watertown, MA, USA) as follows: 2 min at 94 8Cfol- clone us to suspect an ancient introduction of A. donax lowed by 35 cycles of 94 8C for 1 min, 56 8C annealing for 1 min, to the Mediterranean Basin. Here we test the hypothesis of a and 72 8C for 2 min, followed by a final extension of 72 8C for founder effect linked to this introduction from Asia to the 5 min. Purification and sequencing of PCR products were Mediterranean using plastid DNA genetic markers. Secondly, carried out by Eurofins MWG Operon (Ebersberg, Germany). we characterize and localize the invasive lineage among the The five plastid DNA regions were manually aligned in Eurasian distribution of A. donax using genetic, morphometric MEGA 5.05 (Tamura et al., 2011). Haplotype relationships and production data. Finally, we develop a bioclimatic were inferred using the median-joining network algorithm species distribution model calibrated on Mediterranean occur- implemented in Network 4.6(Bandelt et al., 1999). Insertion/ rence data to localize potential Asian origins for the worldwide deletion (indel) sites (including repetitive regions as mini- and invasive genotype. Due to difficulties in obtaining fresh speci- microsatellites) were reduced, considered as a fifth state and mens across the Asian range of A. donax, our sampling is weighted as one-tenth of a substitution. This down-weighting based on herbarium specimens from European collections. In is classicallyadopted forhypervariablesites with high probabilities Hardion et al. — Origin of the invasive Arundo donax (Poaceae) 457 of homoplasy (Saltonstall and Lambertini, 2012). In order to dis- of maximum entropy to estimate the most uniform distribution tinguish haplotypes based on substitution patterns from those within the study area given the constraint that the expected value supported by hypervariable sites, the same analysis was also gen- of each environmental predictor variable under this estimated dis- erated without considering indels. tribution matches its empirical average (Phillips et al., 2006). Modeloutputscorrespond toposterior probabilitiesof habitatsuit- ability.Allcombinationsoftypefeatures(linear,quadratic,thresh- Morphometry old and product) were investigated to find the best model with 50 Seven morphometric variables previously used in resolving replicates (cross-validation, training gain threshold of 0.0001). the taxonomyof Mediterranean Arundowere measured on spike- The best model was chosen as the simplest model with the lets collected in the middle part of the panicle: number of flowers highest area under the curve (AUC) value and the lowest standard per spikelet, and lengths of lower and upper glumes, lemma, deviation among the 50 replicates. After calibration on the 1221 palea, lemma awn and hairs (Hardion et al., 2012). As epi- Mediterranean occurrences, this model was then used on 5 dermal structures are highly resolving in grass systematics arc-min cells located from 15 to 508N in latitude to detect suitable (Prat, 1932; Grass Phylogeny Working Group, 2001) and can bioclimatic conditions across Eurasia. also provide information on polyploid events (e.g. in ; Hansen et al., 2007), we also measured three characters under light microscopy (Dialux 20, Leitz, Weltzer, Germany): length RESULTS 4 2 of stomatal guard cells; density (per 10 mm ); and the Phylogeographic structure number of prickles per millimetre of rib line. To avoid damaging herbarium specimens, leaf peels were prepared with After indel reduction, the plastid DNA alignment of 4518 bp clear nail polish, following Hilu and Randall (1984),and contained 77 variable sites, i.e. 44 substitutions and 33 indels in- mounted on slides. All ten variables were measured ten times and cluding ten microsatellites, eight minisatellites and three inver- averaged for each sample. The occurrence of caryopses was also sions (2, 3 and 5 bp). This dataset distinguishes 13 haplotypes noted as a qualitative feature. Morphological data were explored within A. donax (Fig. 1A), which reduces to nine after the using UPGMA hierarchical clustering on Euclidian distances removal of hypervariable sites (Fig. 1B). Based on the branching using the ade4 package in R v.2.15 (Thioulouse et al., 1997; R position of other Arundo spp., A. donax appears to be polyphyl- Development Core Team, 2013), after replacement of the few etic and divided into eastern and western lineages (Figs 1 and missing values by variable means. For each morphological vari- 2A). Haplotype diversity is divided into four biogeographic clus- able, we used a Kruskal–Wallis rank sum to test for significant ters which differentiate along the Himalayas, i.e. Middle-East differences between the three groups identified by UPGMA cluster- (four haplotypes M), Western Himalaya (two haplotypes W), ing, and the coefficient of determination R2 illustrated the propor- Central Himalaya (three haplotypes C) and Eastern Himalaya– tion of variability explained by this clustering. China (four haplotypes E) (Fig. 1A). These patterns are also supported when considering only substitutions, although the number of haplotypes in the Middle East cluster is reduced to Ecological niche modelling one which groups with haplotypes in the Western Himalaya Using ArcGIS 10 (Environmental Sciences Research Institute, cluster (Fig. 1B). Haplotype M1 was found in the 28 samples Redlands,CA,USA),we createda dataset ofA.donax occurrences from the Mediterranean and Irano-Touranian regions (Fig. 2A), collected from the literature, herbaria and databases of and its nearest relatives(M2, M3and M4),which are distinguished Mediterranean countries, including Croatia (http://hirc.botanic. only by hypervariable sites, are located in Afghanistan and hr/fcd/), France (www.silene.eu), ( Hellenica data- Pakistan (Indus valley). The haplotype M1 was also found in base; Strid, 2000), Palestine (http://flora.huji.ac.il) and Spain invasive samples from New Caledonia, Peru and the USA. (www.anthos.es). To reduce spatial bias due to differential search intensities across regions (Segurado et al., 2006), we sub- Morphometric differentiation sampled this dataset to a grid of 5 arc-min cells (i.e. 0.1 × 0.18) with random origin, reducing the dataset from 3429to 1221occur- The UPGMA hierarchical clustering of morphological char- rences. Based on our biological knowledge of A. donax,wechose acters also divided A. donax into two clusters along a longitudin- five bioclimatic variables from the BioClim database as predica- al gradient (Fig 2A, B): (1) the Mediterranean and Persian tors of its distribution (www.worldclim.org; Hijmans et al., morphotype T1 corresponding to A. donax sensu stricto,with 2005):minimum temperature of the coldest month(BIO6) reflect- the largest morphometric sizes and exclusively associated with ing potentially lethal frost events; mean temperature of the plastid DNA haplotype M1; and (2) the smaller Sino-Himalayan warmest (BIO10) and coldest (BIO11) quarters, delimiting morphotype T2 (Fig. 3). The larger stomatal guard cells of phenological stage for biomass production; precipitation of the the morphotype T1 (mean 38.4 + 4.3 mm) significantly differ driest month (BIO14) indicating potentially lethal from those of the morphotype T2 (mean 29.5 + 3.8 mm) and events; and precipitation of the warmest quarter (BIO18) provid- A. formosana (mean 25.5 + 2.5 mm; Figs 3 and 4). In addition, ing information on water availability during the main period of theleafepidermisofSino-HimalayanmorphotypeT2ischaracter- biomass production. ized by numerous prickles and long hairs, whereas those of mor- To model the potential area of origin for Mediterranean photype T1 show few prickles and not any long hairs (Fig. 4). A. donax, we used a machine learning algorithm based on Although the majority of samples were collected during the presence-only data and implemented in MaxEnt software months (i.e. flowering period), seed-set was found in v. 3.3.3 (Phillips and Dudik, 2008). This method usesthe principle only 12 samples of the morphotype T2 from the Middle-East 458 Hardion et al. — Origin of the invasive Arundo donax (Poaceae)

A (.0.5) south of the Caspian Sea, in southern Iran and along the Indus Valley (Fig. 2C). A. formosana C1

C3 C2 DISCUSSION

W2 Genetic uniformity in the Mediterranean Previous work using AFLPs has questioned the native status of E4 Arundo donax in the Mediterranean, due to its genetic uniformity and lack of seed production (Hardion et al., 2012). Similarly, this E3 study finds no genetic variation in Mediterranean A. donax across W1 several plastid DNA loci and, further, we document the expanded E1 range of this clone to the Middle East and other recently invaded E2 M3 regions worldwide. To date, such an absence of genetic diversity in plastid and nuclear markers has rarely been shown for such a widespread plant species. Based on equivalent markers and sam- M1 M4 pling, the nearest plant model is represented by Pinus pinea, A. micrantha which has a single Mediterranean-wide haplotype and three A. plinii other locally restricted haplotypes (Vendramin et al., 2008). M2 However, those results were explained not only by plant clonality A. donaciformis but also by demographic bottlenecks linked to human-mediated B dispersal. Such patterns of genetic uniformity are more common >4 in invasive species, such as Spartina anglica and Pennisetum 3 setaceum (Poaceae), which also display clonal reproduction, Central 1 polyploid genomes and low genetic diversity (Baumel et al., Himalaya 2001;LeRoux et al., 2007). Despite low mutation rates and pre- ponderant clonality in Arundo (Hardion et al., 2012), our results highlight haplotype variation and phylogeographic structure across the Asian range of A. donax. Consequently, human- mediated dispersal and its invasive ability remain prevailing East explanations for the widespread Mediterranean distribution of Himalaya this non-fruiting clone.

West Himalaya China Restricted origin of a worldwide invasive clone The nearest relatives of the invasive haplotype M1 are found in Middle Afghanistan and Pakistan, along the Indus Valley (haplotypes East M2, M3 and M4). These phylogenetic affinities are reinforced by sequence differences seen only in hypervariable sites (i.e. F IG.1. (A) Combined plastid DNA network of the genus Arundo based on sub- mini- and microsatellites), with mutation rates approx. 105-fold stitutions and mini-/microsatellites, and (B) on substitutions only. Haplotypes are colour coded; their character code corresponds to Supplementary Data Table S1; higher than plastid DNA substitutions (Cozzolino et al., 2003). their size correspond to the number of individuals detected in the dataset. Small Further, species distribution modelling predicts suitable biocli- black circles indicate haplotypes not detected in the dataset. matic conditions for the Mediterranean clone along the Indus Valley and in south-western Iran, where were found, but also to the south of the Caspian Sea. Consequently, this study (south-western Iran, Pakistan and Afghanistan) and along the clearly supports a Middle East origin for the Mediterranean Himalayas (Fig. 2A). clone of A. donax. However, the robust morphotype T1 contrasts with the morphotype T2 which is also found in this region. This mismatch between morphometric and phylogenetic cluster- ing could be a consequence of gigantism associated with differ- Ecological niche modelling ences in ploidy level from fruiting lower cytotypes, as seen Based on 1221 occurrences, the ecological niche model for between A. plinii and A. donaciformis in the Mediterranean A. donax in the Mediterranean projected high importance of bio- (Hardion et al., 2012). This polyploid differentiation is sup- climatic variables BIO10 (mean temperature of the warmest ported by stomatal sizes, which clearly distinguish the two mor- quarter), BIO14 (precipitation in the driest month) and BIO6 photypes within A. donax, probably corresponding to its two (minimum temperature of the coldest month). The most parsimo- ploidy levels: 2n ¼ 18x approx. 108 in the Mediterranean nious model was obtained using only linear features (test AUC (Gorenflot et al., 1972; Hardion et al., 2011, 2013) and 2n ¼ value ¼ 0.90, AUC standard deviation ¼ 0.0065, gain threshold 12x approx. 72 in Asia (Larsen, 1963; Christopher and reached after 180 iterations). Projected on Asian bioclimatic Abraham, 1971; Bochantseva, 1972; Mehra and Kalia, 1975). data, this model suggests the highest occurrence probabilities Following this hypothesis, the seed production of the morphotype Hardion et al. — Origin of the invasive Arundo donax (Poaceae) 459

A Caspian Sea * * Iran Himalayas * * Pakistan * B A. donax T2 * * China Indus * Saudi * A. formosana Valley India A. donax T1 Arabia C

0 0·22 0·44 0·66 0·88 500 km

F IG. 2. (A) Geographical distribution of plastid DNA haplotypes and morphotypes. White circles, A. donax morphotype T1; grey circles, A. donax morphotype T2; black circles, A. formosana morphotype; *, seed occurrence. Coloured rings correspond to plastid DNA haplotypes (Fig. 1). (B) UPGMA based on morphological data. (C) Ecological niche modelling of A. donax calibrated on 1221 Mediterranean occurrences (black dots) and projected on sub-tropical Eurasia using MaxEnt.

350 820 70 360 830 70 36083070 350 830 70 350 820 70 14 7 5 14 14 2 2 2 2 R = 0·647*** R = 0·655*** R = 0·462*** R = 0·681*** 12 12 6 4 12 2 R = 0·429*** 10 10

L 5 pa G1 G2

nbF 10 3 8 8 4 8 2 6 6 3 4 6 4 1 A.d.T1 A.d.T2 A.f. A.d.T1 A.d.T2 A.f. A.d.T1 A.d.T2 A.f. A.d.T1 A.d.T2 A.f. A.d.T1 A.d.T2 A.f.

35082070 350 82070 370 800 70 370 800 70 370 800 70 8 55 5 25 8 2 2 R = 0·253*** 7 50 R = 0·529*** 2 2 R = 0·344*** R = 0·235*** 2 20 R = 0·698*** 4 6 6 45

40

5 A X 15 pL 3 dX dP 4 35 4 10 2 30 2 3 25 1 5 2 0 A.d.T1 A.d.T2 A.f. A.d.T1 A.d.T2 A.f. A.d.T1 A.d.T2 A.f. A.d.T1 A.d.T2 A.f. A.d.T1 A.d.T2 A.f.

F IG. 3. Boxplots of morphological variables between A. donax T1 (A.d.T1), A. donax T2 (A.d.T2) and A. formosana (A.f.), i.e. number of flowers per spikelet (nbF), and length (mm) of lower glume (G1), upper glume (G2), lemma (L), palea (pa), lemma hair (pL) and awn (A), length of stomatal guard cells (X; mm), stoma density (dX; per 104 mm2), numberof rib prickles per mm (dP, per mm of rib line). Grey numbers, numbers of measures per variable per morphotype; bold lines, median values; dashed lines, entire variable range; R2, coefficient of determination; ***Kruskal–Wallis P-value ,0.0001. 460 Hardion et al. — Origin of the invasive Arundo donax (Poaceae)

A C ancient introduction from eastern to western Eurasia. The three datasets used here suggest that this archaeophyte, i.e. alien P species introduced before 1500 AD, originates from the Middle East and was probably introduced to the Mediterranean P Basin in antiquity. In recent times, it has also become a as it has subsequently been introduced around the world to other X locations with similar bioclimatic conditions. Due to its high X level of invasiveness and continued spread worldwide, this clone may represent one of the oldest and most persistent bio- logical invasions. Because human selection and polyploid dif- 50 mm 50 mm ferentiation could have played a crucial role in the current range of A. donax, further studies dealing with phylogeographic, B D cytogenetic and archaeological aspects of its distribution should resolve the human-mediated and evolutionary history of this clonal reed species across Eurasia.

m m 5 m5 m SUPPLEMENTARY DATA Supplementary data are available online at www.aob.oxford F IG. 4. Scanning electron micrographs of the limb abaxial epidermis and stoma of A. donax morphotype T1, with large stoma (X) and few indumentum elements journals.org and consist of Table S1: plant sampling list, geo- (A and B, MDo1), and A. donax morphotype T2, with smaller stoma (X), numer- graphical origin, herbarium barcode, plastid DNA haplotype ous prickles (P) and long hairs (D and C, EDo6). and morphotype assignments. Table S2: GenBank accession numbers. T2 could be linked to its lower polyploidy level of 2n ¼ 12x approx. 72. Further studies including extensive sampling of living material across the Middle East are needed to distinguish ACKNOWLEDGEMENTS closer fruiting relatives from the invasive clone using chromo- The authors thank the people and institutions who helped us to some counts and highly variable molecular markers. obtain specimen loans, plant samples and occurrence data, in- cluding F. Fleury (MARS Marseille, France), T. Haevermans (MNHN Paris, France), L. Gautier and N. Fumeaux (CJB One of the oldest invasive species? Geneva, Switzerland), M. Vorontsova (RBG Kew, UK), The use and trade of reeds by human civilizations since the S. Cubey and A. Smith (RBG Edinburgh, UK), R. Prakash Neolithic has been documented across scholarly fields (NHM London, UK), E. Vitek (NHM Vienna, Austria), (Ta¨ckholm and Drar, 1973; Postgate, 1980; Faiella, 2005) and U. Starck and R. Vogt (BGBM Berlin, Germany), B. Gambin, in numerous Greek and Roman texts summarized by Pliny the F. Abdel-Samad, A. Fridlender and F. Me´dail (Aix-Marseille Elder. For example, several Sumerian tablets mentioned thou- University, France), H. Akhani (Tehran University, Iran), sands of bundles of reed culms being imported across H. Engels, B. Zehzad and C. Constantinou, and also A. Danin Mesopotamia since 2500 BC (Joannes et al., 2001). Among (Flora of Israel, Israel), L.M. Domingo (Anthos, Spain), the reeds, A. donax is an easily transplantable and highly product- T. Nikolic (Flora Croatica, Croatia), V. Noble (Silene, France) ive species offering an exceptional combination of culm robust- and A. Strid (Flora Hellenica, Greece). The authors are also ness, lightness and flexibility (Perdue, 1958). Numerous plant grateful to M. Juin and N. Duong (IMBE) for their help in mo- species have been dispersed by humans from south-western lecularexperiments, and to M. Fayand two anonymous reviewers Asia to the Mediterranean since the early Neolithic (Zohary for their relevant comments. This work is included in an et al., 2012). As a consequence, phylogeographic studies on Interdisciplinary Program of Research Urban-Environment European domesticated species often exhibit phylogenetic (PIRVE) financed by the CNRS, the Ministry of Ecology origin or diversity in the East, mainly from the Levant, e.g. (France) and the DREAL PACA. The authors are also grateful olive (north Levant, Besnard et al., 2013) and for other financial support provided by the Research Federation (Kilian et al.,2007), but also from the Middle East, e.g. grapevine ECCOREV and the Socie´te´ Botanique de France. (Caspian region, Arroyo-Garcia et al., 2006) or the wheat D-genome (Aegilops tauschii, north-eastern Iran, Saeidi et al., 2008), and Central Asia, e.g. domesticated apple (Cornille LITERATURE CITED et al., 2012). Joining the list of ancient introductions from south- Ahmad R, Liow PS, Spencer DF, Jasieniuk M. 2008. Molecular evidence for a western Asiato the Mediterranean, the invasive clone ofA. donax single genetic clone of invasive Arundo donax in the . Aquatic Botany 88: 113–120. could even represent one of the oldest plant invasions, in accord- Arroyo-Garcı´a R, Ruiz-Garcı´a L, Bolling L, et al. 2006. Multiple origins of ance with its broad naturalized distribution. cultivated grapevine (Vitis vinifera L. ssp. sativa) based on chloroplast DNA polymorphisms. Molecular Ecology 15: 3707–3714. Balogh E, Herr JMJr, Czako´ M, Ma´rton L. 2012. Defective development of Conclusions male and female in Arundo donax L. (Poaceae). Biomass and Bioenergy 45: 265–269. The present study confirms the genetic uniformity of A. donax Bandelt HJ, Forster P, Ro¨hl A. 1999. Median-joining networks for inferring from the Mediterranean to Iran, supporting the hypothesis of its intraspecific phylogenies. Molecular Biology and Evolution 16: 37–48. Hardion et al. — Origin of the invasive Arundo donax (Poaceae) 461

Baumel A, Ainouche ML, Levasseur JE. 2001. Molecular investigations in Joanne`s F, Michel C, Bachelot L. 2001. Dictionnaire de la civilisation me´sopo- populations of Spartina anglica C.E. Hubbard (Poaceae) invading coastal tamienne. Paris: R. Laffont. Brittany (France). Molecular Ecology 10: 1689–1701. Johnson M, Dudley T, Burns C. 2006. Seed production in Arundo donax. Baumel A, Ainouche M, Kalendar R, Schulman AH. 2002. Retrotransposons Cal-IPC News 14: 12–13. and genomic stability in populations of the young allopolyploid species Kilian B, O¨ zkan H, Deusch O, et al. 2007. Independent wheat B and G genome Spartina anglica C.E. Hubbard (Poaceae). Molecular Biology and origins in outcrossing Aegilops progenitor haplotypes. Molecular Biology Evolution 19: 1218–1227. and Evolution 24: 217–227. Besnard G, Khadari B, Navascue´sM,et al. 2013. The complex history of the Lafuma L, Balkwill K, Imbert E, Verlaque R, Maurice S. 2003. Ploidy level olive tree: from Late Quaternary diversification of Mediterranean lineages and originof the Europeaninvasiveweed Senecio inaequidens (Asteraceae). to primary domestication in the northern Levant. Proceedings of the Plant Systematics and Evolution 243: 59–72. Royal Society B: Biological Sciences 280: 2012–2833. Larsen K. 1963. Studies in the flora of Thailand 14. Cytological studies in vas- Blight O, Berville L, Vogel V, et al. 2012. Variation in the level of aggression, cular of Thailand. Dansk Botanisk Arkiv 20: 211–275. chemical and genetic distance among three supercolonies of the La Sorte FA, Pysˇek P. 2009. Extra-regional residence time as a correlate of plant Argentine ant in Europe. Molecular Ecology 21: 4106–4121. invasiveness: European archaeophytes in North America. Ecology 90: Bochantseva ZP. 1972. Ochislakh chromosom. Introd. Akkli. Rasteny 2589–2597. Akademiya Nauk Uzbekskoi SSR: 44–53. Le Roux JJ, Wieczorek AM, Wright MG, Tran CT. 2007. Super-genotype: Bor NL. 1970. Gramineae. In: Rechinger KH, ed. Flora Iranica 70: 147–244. global monoclonality defies the odds of Nature. PLoS One 2: e590. Brach AR,SongH. 2006.eFloras:new directionsforonlinefloras exemplified by Mariani C, Cabrini R, Danin A, et al. 2010. Origin, diffusion and reproduction the Flora of China Project. Taxon 55: 188–192. of the giant reed (Arundo donax L.): a promising weedy . Annals Bucci A, Cassani E, Landoni M, Cantaluppi E, Pilu R. 2013. Analysis of of Applied Biology 157: 191–202. chromosome number and speculations on the origin of Arundo donax Mehra PN, Kalia V. 1975. IOPB chromosome number reports XLIX. Taxon 24: L. (Giant Reed). Cytology and Genetics 47: 237—241. 501–516. Carlton JT. 1996. Biological invasions and cryptogenic species. Ecology 77: Perdue RE. 1958. Arundo donax, source of musical reeds and industrial cellu- 1653–1655. lose. Economic Botany 12: 368–404. Christopher J, Abraham A. 1971. Studies on the cytology and phylogeny of Phillips SJ, Dudı´k M. 2008. Modeling of species distributions with Maxent: new south Indian grasses: 1. Subfamilies Bambusoideae, Oryzoideae, extensions and a comprehensive evaluation. Ecography 31: 161–175. and Festucoideae. Cytologia 36: 579–594. Phillips SJ, Anderson RP, Schapire RE. 2006. Maximum entropy modeling of Clay K. 2003. Conservation biology: parasites lost. Nature 421: 585–586. species geographic distributions. Ecological Modelling 190: 231–259. Cornille A, Gladieux P, Smulders MJ, et al. 2012. New insight into the history Pimentel D, Zuniga R, Morrison D. 2005. Update on the environmental and of domesticated apple: secondary contribution of the European wild apple to economic costs associated with alien-invasive species in the United the genome of cultivated varieties. PLoS Genetics 8: e1002703. States. Ecological Economics 52: 273–288. Pilu R, Cassani E, Landoni M, et al. 2014. Cozzolino S, Cafasso D, Pellegrino G, Musacchio A, Widmer A. 2003. Genetic characterization of an Italian giant reed (Arundo donax L.) clones collection: exploiting clonal selection. Fine-scale phylogeographical analysis of Mediterranean Anacamptis palus- Euphytica 196: 169–181. tris (Orchidaceae) populations based on chloroplast minisatellite and micro- Pizzolongo P. 1962. Osservazioni cariologiche su Arundo donax e Arundo plinii. satellite variation. Molecular Ecology 12: 2783–2792. Annuali di Botanica (Roma) 27: 173–187. Dandelot S, Verlaque R, Dutartre A, Cazaubon A. 2005. Ecological, dynamic Postgate JN. 1980. Palm-trees, reeds and rushes in Iraq ancient and modern. In: and taxonomic problems due to Ludwigia (Onagraceae) in France. Barrelet MT, ed. L’arche´ologie de l’Iraq du de´but de l’e´poque Ne´olithique a` Hydrobiologia 551: 131–136. 33 av. notre e`re. Paris, France: CNRS editions: 99–109. Dlugosch KM, Parker IM. 2008. Founding events in species invasions: genetic Prat H. 1932. L’e´piderme des Gramine´es. Etude anatomique et syste´matique. variation, adaptive evolution, and the role of multiple introductions. Annales des Sciences Naturelles, Botanique 14: 117–324. Molecular Ecology 17: 431–449. Preston CD, Pearman DA, Hall AR. 2004. Archaeophytes in Britain. Botanical Doyle J, Doyle JL. 1987. A rapid DNA isolation procedure for small quantities of Journal of the Linnean Society 145: 257—294. fresh leaf . Phytochemical Bulletin, Botanical Society of America 19: Pysˇek P, Richardson DM. 2006. The of naturalization in alien 11–15. plants. Journal of Biogeography 33: 2040–2050. Elton CS. 1958. The ecology of invasions by animals and plants. Chicago: Pysˇek P, Richardson DM, Rejma´nek M, Webster GL, Williamson M, University of Chicago Press. Kirschner J. 2004a. Alien plants in checklists and floras: towards better Faiella G. 2005. The technology of Mesopotamia. New York: The Rosen communication between taxonomists and ecologists. Taxon 53: 131–143. Publishing Group. Pysˇek P, Richardson DM, Williamson M. 2004b. Predicting and explaining Gorenflot R, Raicu P, Cartier D. 1972. Caryologie de la canne de Provence plant invasions through analysis of source area floras: some critical consid- (Arundo donax L.). Comptes Rendus de l’Acade´mie des Sciences, Paris erations. Diversity and Distributions 10: 179–187. (Sciences Naturelles) 274: 391–393. PysˇekP, Jarosˇ´ık V, Chytry´ M,Kropa´c Z, Tichy´ L, Wild J.2005. Alienplantsin Grass Phylogeny Working Group. 2001. Phylogeny and subfamilial classifica- temperateweed communities: prehistoric and recent invaders occupy differ- tion of the grasses (Poaceae). Annals of the Missouri 88: ent habitats. Ecology 86: 772–785. 373–457. R Development Core Team. 2013. R: a language and environment for statistical Hansen DL, Lambertini C, Jampeetong A, Brix H. 2007. Clone-specific dif- computing. R Foundation for Statistical Computing, Vienna, Austria. ferences in Phragmites australis: effects of ploidy level and geographic Sablok G, Fu Y, Bobbio V, et al. 2014. Fuelling genetic and metabolic explor- origin. Aquatic Botany 86: 269–279. ation of C3 bioenergy crops through the first reference transcriptome of Hardion L, Verlaque R, Fridlender A, Vila B. 2011. IAPT/IOPB chromosome Arundo donax L. Plant Biotechnology Journal 12: 554–567. data 11. Taxon 60: 1221. Saeidi H, Rahiminejad MR, Heslop-Harrison JS. 2008. Retroelement inser- Hardion L, Verlaque R, Baumel A, Juin M, Vila B. 2012. Revised systematics tional polymorphisms, diversity and phylogeography within diploid, of Mediterranean Arundo (Poaceae) based on AFLP fingerprints and morph- D-genome Aegilops tauschii (Triticeae, Poaceae) sub-taxa in Iran. Annals ology. Taxon 61: 1217–1226. of Botany 101: 855–861. Hardion L, Verlaque R, Fridlender A, Zehzad B, Vila B. 2013. IAPT/IOPB Saltonstall K. 2001. A set of primers for amplification of noncoding regions of chromosome data 15. Taxon 62: 1075. chloroplast DNA in the grasses. Molecular Ecology Notes 1: 76–78. Herrera AM, Dudley TL. 2003. Reduction of riparian arthropod abundance and Saltonstall K. 2002. Cryptic invasion of a non-native genotype of the common diversity as a consequence of giant reed (Arundo donax) invasion. reed, Phragmites australis, into North America. Proceedings of the Biological Invasions 5: 167–177. National Academy of Sciences, USA 99: 2445–2449. Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A. 2005. Very high reso- Saltonstall K, Lambertini C. 2012. The value of repetitive sequences in chloro- lution interpolated climate surfaces for global land areas. International plast DNA for phylogeographic inference: a comment on Vachon and Journal of Climatology 25: 1965–1978. Freeland 2011. Molecular Ecology Resources 12: 581–585. Hilu KW, Randall JL. 1984. Convenient method for studying grass leaf epider- Segurado P, Arau´jo MB, Kunin WE. 2006. Consequences of spatial autocorrel- mis. Taxon 33: 413–415. ation for niche-based models. Journal of Applied Ecology 43: 433–444. 462 Hardion et al. — Origin of the invasive Arundo donax (Poaceae)

Shea K, Chesson P. 2002. Community ecology theory as a framework for bio- Thioulouse J, Chessel D, Dole S, Olivier J-M. 1997. ADE-4: a multivariate ana- logical invasions. Trends in Ecology and Evolution 17: 170–176. lysis and graphical display software. Statistics and Computing 7: 75–83. Strid A. 2000. Remove from marked Records The Flora Hellenica database. Vendramin GG, Fady B, Gonza´lez-Martı´nez SC, et al. 2008. Genetically de- Portugaliae Acta Biologica 19: 49–59. pauperate but widespread: the case of an emblematic Mediterranean pine. Suehs CM, Affre L, Me´dail F. 2004. Invasion dynamics of two alien Evolution 62: 680–688. Carpobrotus (Aizoaceae) taxa on a Mediterranean island: II. Verlaque M, Durand C, Huisman J, Boudouresque C-F, Le Parco Y. 2003. Reproductive strategies. Heredity 92: 550–556. On the identityand originof the Mediterranean invasive Caulerpa racemosa Taberlet P, Gielly L, Pautou G, Bouvet J. 1991. Universal primers for amplifi- (Caulerpales, Chlorophyta). European Journal of 38: 325–339. cation of three non-coding regions of chloroplast DNA. Plant Molecular Verlaque R, Affre L, Diadema K, Suehs CM, Me´dail F. 2011. Unexpected Biology 17: 1105–1109. morphological and karyological changes in invasive Carpobrotus Ta¨ckholm V, Drar M. 1973. Flora of Egypt, Vol. II. Koenigstein: Otto Koeltz (Aizoaceae) in Provence (SE France) compared to native South African Antiquariat. species. Comptes Rendus de Biologie 334: 311–319. Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S. 2011. Wijte AHBM, Mizutani T, Motamed ER, Merryfield ML, Miller, Alexander MEGA5: Molecular Evolutionary Genetics Analysis using maximum like- DE. 2005. Temperature and endogenous factors cause seasonal patterns in lihood, evolutionary distance, and maximum parsimony methods. rooting by stem fragments of the invasive Giant Reed, Arundo donax Molecular Biology and Evolution 28: 2731–2739. (Poaceae). International Journal of Plant Sciences 166: 507–517. Telle S, Thines M. 2008. Amplification of cox2 (620 bp) from 2 mg of up to Zohary D, Hopf M, Weiss E. 2012. Domestication of plants in the Old World:the 129 years old herbarium specimens, comparing 19 extraction methods origin and spread of domesticated plants in southwest Asia, Europe, and the and 15 polymerases. PLoS One 3: e3584. Mediterranean Basin. Oxford: Oxford University Press.